Harmonization of Four PET Scanners Through the Use of a Uniformity Phantom and a Sphere Simulation Tool
D Gomez Cardona1*, J Browne2, L Stecker3, R Pooley3, (1) Gundersen Health System, La Crosse, WI, (2) Mayo Clinic, Rochester, MN, (3) Mayo Clinic, Jacksonville, FL
Presentations
SU-C-TRACK 6-2 (Sunday, 7/25/2021) 1:00 PM - 2:00 PM [Eastern Time (GMT-4)]
Purpose: Quantitative positron emission tomography (PET) with 18F-fluorodeoxyglucose (18F-FDG) can provide important information for staging cancer and monitoring therapy response. However, to reduce quantitative bias and variability, hospitals should standardize their imaging procedures and harmonize the SUV response of their fleet of PET scanners. The aim of this study was to harmonize the response of four PET scanners in our practice through the use of a uniformity phantom and a sphere simulation tool which predicts optimal reconstruction filter settings[1]. The predicted reconstruction settings were validated experimentally using the NEMA NU2-2007 phantom.
Methods: Four PET systems, GE Discovery MI, GE Signa PETMR, Siemens Biograph 40, and Siemens Biograph Vision were assessed. The total activity of 18F-FDG in the uniformity and NEMA phantoms was 70MBq; the NEMA phantom spheres-background ratio was 20:2kBq/ml. The uniformity and NEMA phantoms were scanned for 20 min/bed position (1-3 bed positions) and 2 min/bed position (1 -2 bed positions), respectively. Two clinical reconstruction settings were evaluated for each scanner to identify baseline settings. The optimal post-filter settings were predicted using the uniformity phantom and the sphere simulation tool. These settings were applied to the NEMA phantom data to validate the sphere simulation tool. For each sphere in the NEMA phantom, the maximum voxel value and an adaptive 50% isocontour mean value were calculated[2].
Results: The recovery coefficient (RC) curves obtained for the NEMA and uniformity phantom show close agreement (Figure 1a and 1b). Figure 1b also displays a harmonized RC curve after specific filter settings were applied. Figure 2a and 2b display pre- and post-harmonization results of the NEMA phantom following the application of the predicted filter settings.
Conclusion: Scanner harmonization can be achieved using vendor specific uniformity phantoms and the sphere simulation tool rather than the more expensive and complex NEMA phantom.
Handouts
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email